Leads research on Cursor's agentic coding model Composer 2, focusing on model training strategy, RL methodology, and specialization for software engineering within the Cursor environment.
no scored calls yet — needs a stated position or a categorical verdict, with a matured window vs SPY
Cursor argues that allocating all model weights to a single task (software engineering inside Cursor) enables order-of-magnitude lower inference cost vs general models like Opus, while achieving superior performance through RL on their proprietary harness and user data.
Cursor selected Kimi 2.5 (1T parameter MoE with 30B active parameters) as the base model for Composer 2, leveraging its sparse architecture and strong code capabilities before applying continual pre-training and large-scale RL.